#23097: The circuit timeout prediction is not working properly
--------------------------+------------------------------------
Reporter: dgoulet | Owner:
Type: defect | Status: new
Priority: Very High | Milestone: Tor: 0.3.2.x-final
Component: Core Tor/Tor | Version:
Severity: Normal | Resolution:
Keywords: 031-backport | Actual Points:
Parent ID: | Points:
Reviewer: | Sponsor:
--------------------------+------------------------------------
Comment (by dgoulet):
Replying to [comment:3 mikeperry]:
[...]
> One option is to call rep_hist_note_used_internal() from the
needs_hs_server_circuits() block in circuit_predict_and_launch_new(),
since it seems like we always want to be predicting hidden service
circuits continuously if an onion service is configured?
I'm actually puzzled. Right now, we assume that we need to have 3 internal
circuits at all time for the service
(`SUFFICIENT_UPTIME_INTERNAL_HS_SERVERS 3`) but I don't think that is
optimal.
First, we now have to consider 2 HS subsystem (v2 and v3). Second, a
service doesn't really know how many circuits it needs because the number
of clients connecting is arbitrary that is can be 1 once a day or 1000
every minute. In terms of IP circuit, we open 3 early and keep them for a
while so prediction won't help us much imo.
Bottom line, the current value of `3` is kind of "whatever why?" and we
should only care about RP circuits meaning `need_capacity = 1` always.
That being said, for now and short term, I think we should do what you
propose, in `needs_hs_server_circuits()` and always consider that we need
capacity meaning: `rep_hist_note_used_internal(now, 1, 1)`.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23097#comment:4>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@lists.torproject.org
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs
